The present application relates to a method, machine and computer program product to provision, configure and manage a computer network, and more specifically, to provision, configure and manage a wavelength division multiplexed computer network using a virtual command interface.
A three-dimensional virtual environment providing near real-time, streaming visual representations for data center management has been disclosed in U.S. patent application Ser. No. 11/747,157 entitled VIRTUAL NETWORK WDM NETWORK 170 filed 10 May 2007, and U.S. patent application Ser. No. 11/747,147 filed on 10 May 2007 entitled HOLOGRAPHIC ENTERPRISE NETWORK, both of which are hereby incorporated by reference in their entirety and both have as a common assignee the assignee of this patent.
Hardware elements such as servers, racks, and power and cooling are structurally organized and visually represented in the virtual command center that displays platform(s) for equipment, observation decks and catwalks, display screens, and various infrastructures such as the in-world communications gear. The application of virtual command centers, however, has not been extended to wavelength division multiplexing networks within local area networks, metropolitan area networks, and wide area networks.
In fiber optic communications, wavelength division multiplexing (hereinafter WDM) is a technology which multiplexes multiple optical signals on a single optical fiber using different wavelengths of laser light to carry different signals. A wavelength division multiplexing system uses a multiplexer at the transmitter to join the different wavelength signals together and a demultiplexer at the receiver to split them apart. Because fiber optic communication channels share a common physical path, wavelength division multiplexing expands the capacity of a network without laying more optical fibers which in turn reduces the cost of leased optical fiber while enabling greater communication between remote data centers. In fact, an existing optical infrastructure of WDM and optical amplifiers is capable of accommodating several generations of technology development without having to change the backbone network. By simply upgrading the multiplexers and demultiplexers at each end, the capacity of a network link is expanded.
A WDM network is often employed in a metropolitan area network (MAN). WDM networking equipment has unique and difficult management challenges. First, WDM networking equipment does not automatically go onto the networks and discover the types of channels, e.g., Ethernet, Infiniband, Fibre Channel, PCI Express, Serial ATA, VoIP, connected on the network. Second, the WDM networking equipment must be manually provisioned and commissioned by a service technician prior to use. Moreover, whenever a channel type is modified, the WDM networking equipment must be manually provisioned and commissioned again. A WDM network requires networking equipment at each location; the locations, moreover, are typically quite far apart, so that at least two technicians are required to be at the different locations on the WDM network at the same time, communicating via cell phone or other means to coordinate settings on both ends of a link or a point-to-point connection on the WDM network. If the settings on both ends of the link are not the same, the link does not operate properly.
Most WDM networks use Transaction Language 1 (TL1) messages for management of the network resources. TL1 comprises a set of ASCII-based instructions or “messages” to manage a network element and its resources. Network element vendors use TL1 messages to implement embedded management interfaces for their network elements. TL1 is the dominant management protocol for controlling telecommunications networks in North America today and service providers have large TL1-based management operation support systems that control the network.
In any network system, it is critical to validate the proper configuration of the network equipment and enforce configuration constraints required by the servers. As mentioned, provisioning an existing network to have a new channel or a new element requires a human being to go to the equipment and manually configure and provision it. Examples of configuration constraints are that a FICON (Fiber Connectivity) channel should not be connected to a network interface that doesn't support the FICON data rate; another is that two FICON channels that are supposed to be redundant actually should be provisioned across two physically redundant WDM paths. Improper configuration of a WDM client interface to a server can result in a failed link and/or significantly more bit errors to which the server will respond by retransmitting numerous requests consuming the server's resources. Unless a redundant or high availability configuration is validated the system administrator could mistakenly think that the network is fault tolerant. In order to correct a problem that arises, moreover, the administrator may attempt to vary a path offline for maintenance, not realizing she/he is disrupting all connectivity to the remote location. It is thus important to validate the configuration of a WDM network.
Once configured, among the unique and most important requirements of a WDM network interface is protection switching. The WDM equipment offers protection switching that splits traffic across two physically redundant paths, and when one path is detected as being interrupted, the WDM equipment automatically switches data to the redundant or backup path. Protection switching may be implemented as a unidirectional path switched ring or a bidirectional line switched ring, provisioned within the network equipment. The faillover switch usually occurs within 50-100 milliseconds or longer.
Sometimes it becomes necessary to provision a link offline for maintenance, such as repair of a degraded optical fiber connection. To date, this reprovisioning is achieved manually. Thus, dynamic switching, such as switching from a longer path to a shorter path to reduce latency between the attached servers is necessary. Such dynamic switching will improve performance and/or reduce the need for buffer credit flow control under some protocols such as FICON or Fibre Channel.
Protection switching as above is required for disaster recovery and other synchronous applications, but these applications have additional requirements. Timing information of failures or protection switches and other events are not coordinated on the WDM network until equipment logs are taken after the fact and reviewed manually.
The Open Systems Interconnection Reference Model (OSI Reference Model or OSI Model) is an abstract description for layered communications and computer network protocol design that, in its most basic form, divides network architecture into seven layers which, from top to bottom, are the Application, Presentation, Session, Transport, Network, Data-Link, and Physical Layers. There are quality of service monitoring mechanisms for Internet traffic that are supposed to prevent network congestion or delays time through monitoring at various OSI levels but they only monitor internet protocols within a server; these mechanisms do not enable communication between a server and network element. There are, moreover, network management systems that control optical switches using a supervisory communication channel but the switches do not connect the network management system to any other management system, such as another server. Consequently, an attached server is not aware of switching operations within the network. These switches, moreover do not process TL1 commands, correlate events on the server and network, or translate commands between a server and networking equipment.
There is also a fully transparent network switch that reports its status to the network management interface but it is independent of any attached server equipment. While both the switch and network management interface support TL1 protocol and other management interfaces, they do not connect to subtended equipment so that servers connected to the WDM are unaware of switch events in the network. Network elements such as routers, switches, and hubs are capable of providing Simple Network Management Protocol (SNMP) messages for conditions that warrant administrative attention, such as checking of the IP network configuration and IP protocols, but these equipment do not collect information from servers attached to the network or correlate/validate the servers' requirements with the network configuration. These network elements, moreover, do not perform in multiprotocol data center environments.
A virtualized control plane used in a high speed optical network testbed has a configurator tool with a graphical user interface (GUI) interface used to configure network parameters across multiple network elements. These network elements support GMPLS, a wide area networking protocol for wavelength routing. The tool, however, does not collect information from servers attached to the network; nor can it validate server requirements because server platforms do not support GMPLS. A wavelength multiplexer can be provisioned for data protection, performance monitoring, and other features using an SNMP interface option but these existing multiplexers also do not collect or correlate corresponding information from servers attached to the network. An optical provisioning layer in a WDM network currently used to reconfigure the topology and reprovision services in networking equipment has no capability to collect or correlate the corresponding information from servers attached to the network. Currently, bandwidth provisioning on a WDM cannot be accomplished through direct communication between the server and network interfaces. Identification of the number and type of channels required by the server are not done on existing WDMs.
At present, timing of add/drop or other events within a WDM network is not correlated with other events in the servers or other equipment connected on the network. A clock source within the WDM network (compatible with SONET standards) may generate a synchronous clocking signal to retime and regenerate signals after long distances but these signals do not signify or communicate events to any servers attached to the network. Quality of service software that can convert traffic into and schedule IP data packets also does not interface with the attached servers. The timing of network protection switch events using a typical WDM optical switch cannot be controlled by servers transmitting data over the network.
Transmit data queues and receiver queues are used to associate an identity with a given packet of data but they do not correlate the timing of network events with those on a server attached to the network. Tracking a sequence of packets as they transit through a network node as in, for example, logging the time when a given packet passes a given node, is performed relative to a reference clock within the network, but again that time is not correlated with any external server clocks thereby making it impossible to associate transit times within the network and events on attached servers.
Servers on a WDM network transmit and receive TL1 commands but to date, there is no TL1 command translation or encapsulation within a virtual network management interface. TL1 commands are not translated into a format compatible with existing computer management interfaces, or converted to other formats. There is no coordinated management between the server and WDM equipment, particularly for WDM equipment which may be located very far away from the server, e.g., 100 kilometers or more. To date, the network does not respond to changes in network resources or changing conditions on an attached server.
There exists a further need to correlate network events with server processing and messaging.